I was chatting with one or two friends about this idea, and I thought I'd ask you what you think. Basically, can one build an "analogue switch" by using ordinary transistors (FETs or BJTs or CD4066 even) to switch on and switch off the power supply to opamps?
Essentially, imagine a circuit where there's one dual-opamp, set up as unity gain buffer, for each input. Dual opamps can handle left and right channel in one chip, hence dual. The power supply of each opamp is controlled by two analog switches. These switches can be built using transistors/CD4066 type things.
The purpose of doing all this is to get low-distortion analog switches. It is believed that even the old NE5532 is lower distortion than most FET or MOS analog switches. Hence this line of thinking. The aim here is to pass the audio signal through only good opamps, never through the analog switches themselves. For the moment, assume that the user will be willing to tolerate the sub-second audible glitch/noise when one opamp switches off and another switches on, when the input selector is changed.
Please forgive me if this idea is totally ludicrous. I'm too new to all this to know the difference.
Tarun
Essentially, imagine a circuit where there's one dual-opamp, set up as unity gain buffer, for each input. Dual opamps can handle left and right channel in one chip, hence dual. The power supply of each opamp is controlled by two analog switches. These switches can be built using transistors/CD4066 type things.
The purpose of doing all this is to get low-distortion analog switches. It is believed that even the old NE5532 is lower distortion than most FET or MOS analog switches. Hence this line of thinking. The aim here is to pass the audio signal through only good opamps, never through the analog switches themselves. For the moment, assume that the user will be willing to tolerate the sub-second audible glitch/noise when one opamp switches off and another switches on, when the input selector is changed.
Please forgive me if this idea is totally ludicrous. I'm too new to all this to know the difference.
Tarun
To some extent, this would depend on if the topology is inverting or non-inverting. A standard inverting circuit would leak, but the non-inverting should be quite OK.
However, how a signal travels through a non-powered OP-amp is most likely undefined, so it's hard to be sure of a blanket statement. Either test it and work from there, or just regard it as unsupported.
Rune
However, how a signal travels through a non-powered OP-amp is most likely undefined, so it's hard to be sure of a blanket statement. Either test it and work from there, or just regard it as unsupported.
Rune
TI and Intersil
among others have OP-AMPS with "Disable" pins -- I believe that you can search by this parameter in the mfr's database.
a signal will definitely fly through an unpowered opamp, but it won't look so pretty at the output.
I am using some of the OPA3684 current feedback opamp with disable in a DIY phase-meter.
among others have OP-AMPS with "Disable" pins -- I believe that you can search by this parameter in the mfr's database.
a signal will definitely fly through an unpowered opamp, but it won't look so pretty at the output.
I am using some of the OPA3684 current feedback opamp with disable in a DIY phase-meter.
Even if the signal didn't make it through the powered-off op-amp, it would still be like connecting some unbiased PN junctions (ie nonlinear impedances) in parallel with the output. These would cause distortion on the signal that was driven by the one powered op-amp.
Use analog switches or op-amps with disable pins.
Use analog switches or op-amps with disable pins.
So.. instead of gating the power pin, why not short the input to gnd?
For a non-inverting opamp, you can add a series resistance with the source. For an inverting opamp you have a built in to the configuration. You'd still need either an active or passive mixer circuit on the outputs.
I think you could use a tri-state buffer/tranceiver (74LS245 off hand) to short the signals... They should be able to sink the kind of input curren't you're talking about.
--
Danny
For a non-inverting opamp, you can add a series resistance with the source. For an inverting opamp you have a built in to the configuration. You'd still need either an active or passive mixer circuit on the outputs.
I think you could use a tri-state buffer/tranceiver (74LS245 off hand) to short the signals... They should be able to sink the kind of input curren't you're talking about.
--
Danny
well, you can't just nilly willy short the input to ground, you have to look at the circuitry ahead of the input opamp and make sure that by doing so you aren't inadvertently lowering the impedance of the next following circuit.
Op amps are so cheap that I am taken to buffering all my differential signals in instrumentation (or just using an instrumentation amp.)
Get some samples from TI - is my suggestion.
Op amps are so cheap that I am taken to buffering all my differential signals in instrumentation (or just using an instrumentation amp.)
Get some samples from TI - is my suggestion.
jackinnj said:
Sure, but that's why I mentioned adding some input resistance.
If there isn't an EQ circuit (or even if there is, who cares if the source isn't being used) then a pretty normal configuration is to run the input straight into the opamp. Your input impedance for the preceding device is set by the input impedance to the opamp. For an inverting configuration R1 taps to a virtual ground already. For a non-inverting configuration it's good practice to add an input resistor anyway. You won't short your previous stages output by shunting the inputs to gnd.
If there is other stuff on the same signal path before the opamp, I agree then you should consider the impact.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.